Autism has no single cause. It results from a combination of genetic factors, differences in early brain development, and certain prenatal environmental exposures. Genetics plays the dominant role, with heritability estimates ranging from 64% to 92% depending on the study. The CDC’s most recent data, from 2022, puts autism prevalence at about 1 in 31 children aged 8, making it one of the most common neurodevelopmental conditions.
Genetics Is the Strongest Factor
Twin studies consistently show that autism is one of the most heritable conditions in medicine. When one identical twin has autism, the other twin does too roughly 39% to 58% of the time. For fraternal twins, that rate drops to 24% to 31%. A meta-analysis of twin studies placed overall heritability between 64% and 91%. One particularly large study, which analyzed insurance claims covering more than a third of the U.S. population, estimated heritability at approximately 92%, the highest of the 149 diseases and conditions examined.
Siblings of autistic children have about a 10% chance of also being diagnosed, far higher than the general population rate. This familial clustering reinforces how central genetics is to autism risk.
The genetic picture is complex, though. There is no single “autism gene.” Instead, hundreds of genes contribute small amounts of risk. Common genetic variations, the kind everyone carries, collectively account for an estimated 15% to 50% of total autism risk. The contribution is higher in families where multiple members are autistic.
Spontaneous Mutations Play a Major Role
Beyond inherited genetics, spontaneous mutations that appear for the first time in a child (not carried by either parent) are a significant cause. These “de novo” mutations are especially important in families with no prior history of autism. In families where only one person is autistic and there’s no broader family pattern, de novo mutations may account for 52% to 67% of cases. In families where autism already runs in the family, these spontaneous changes contribute far less, around 9% to 11%, because inherited genetic risk is doing most of the work.
This distinction helps explain why autism sometimes seems to “come out of nowhere” in a family. A child can develop autism from genetic changes that neither parent carries, arising spontaneously during the formation of sperm, eggs, or early embryonic development.
How Brain Development Differs
One of the clearest biological differences found in autism involves the brain’s natural process of trimming excess connections between neurons. In typical development, a burst of new connections forms during infancy, especially in the cortex, the brain region involved in social behavior, communication, and sensory processing. By late adolescence, about half of these connections are pruned away, a process that helps the brain become more efficient.
Research from Columbia University found that in children with autism, this pruning process is dramatically slowed. By late childhood, connection density had dropped by about 50% in typical brains but by only 16% in autistic brains. The result is a surplus of neural connections, which may contribute to the sensory sensitivity, information processing differences, and behavioral patterns characteristic of autism.
The researchers traced this pruning defect to an overactive protein that interferes with the brain’s ability to break down and recycle old cellular components. Essentially, the cleanup system that would normally clear the way for pruning isn’t working efficiently. Brain cells in the autistic children studied were filled with old, undegraded parts. This mechanism may represent a common biological pathway across many cases of autism, even those with different genetic origins.
Prenatal Environmental Exposures
While genetics dominates, certain environmental factors during pregnancy can increase risk, likely by interacting with existing genetic vulnerability.
Maternal immune activation, where the mother’s immune system produces significant inflammation during pregnancy, is an established prenatal risk factor. Infections or other sources of inflammation can alter fetal brain development through changes in immune signaling that cross the placenta. The risk appears to vary by the sex of the child, with males generally showing greater vulnerability.
Air pollution exposure during pregnancy and early childhood has also been linked to increased risk. A Harvard study found that for every 10 micrograms per cubic meter increase in fine particulate matter, autism risk rose by 31% during the prenatal period and 64% during early childhood. The third trimester appeared to be the most sensitive window during pregnancy.
Certain medications matter as well. Children exposed to valproate (an anti-seizure medication) in the womb have roughly three times the risk of autism compared to unexposed children. Among children born to mothers with epilepsy, the absolute risk of autism was about 4% with valproate exposure versus about 2.4% without it. This is one reason doctors now carefully weigh the risks and benefits of this medication during pregnancy.
Parental Age and Risk
Advanced paternal age is a well-documented risk factor. A study of over 5.7 million children across five countries found that fathers in their 40s had a 28% higher chance of having an autistic child compared to fathers under 30. For fathers in their 50s, the risk was 66% higher. This is likely because sperm cells accumulate more spontaneous mutations with each passing year, increasing the chance of de novo genetic changes that affect brain development.
Epigenetics: The Layer Between Genes and Environment
Epigenetics helps explain how environmental exposures can influence autism risk without changing DNA itself. Epigenetic modifications are chemical tags attached to DNA or to the proteins that package DNA. These tags act like dimmer switches, turning genes up or down without altering the genetic code. In autism, researchers have found that these switches can be set incorrectly, disrupting gene networks critical for brain development and the formation of connections between neurons.
Both too much and too little of these chemical modifications have been observed in autistic individuals, and the affected genes tend to cluster in pathways important for neurodevelopment. Environmental exposures like certain toxins (including bisphenol A and lead) can alter these epigenetic marks, as can medications like valproate. This creates a plausible bridge between environmental risk factors and the biological changes seen in autism: an exposure during pregnancy doesn’t mutate a gene, but it can change how that gene behaves during critical windows of brain development.
Vaccines Do Not Cause Autism
The claim that vaccines cause autism has been thoroughly investigated and definitively rejected. Three large studies covering populations in the United States, United Kingdom, and Denmark found no association between thimerosal-containing vaccines and autism. A separate study that directly compared thimerosal exposure in children with and without autism found no difference in exposure levels between the two groups.
Perhaps the most telling evidence comes from Denmark and Sweden, which stopped using thimerosal in vaccines in 1992. If thimerosal caused autism, rates should have dropped afterward. Instead, autism rates continued to increase in both countries through at least 1999. The same pattern held in California. These natural experiments across entire national populations confirm that the rise in autism diagnoses is unrelated to vaccine ingredients. The original 1998 study claiming a link was retracted, and its author lost his medical license for ethical violations and data manipulation.